We opensource our **Qwen** series, now including **Qwen**, the base language models, namely **Qwen-1.8B**, **Qwen-7B**, **Qwen-14B**, and **Qwen-72B**, as well as **Qwen-Chat**, the chat models, namely **Qwen-1.8B-Chat**, **Qwen-7B-Chat**, **Qwen-14B-Chat**, and **Qwen-72B-Chat**. Links are on the above table. Click them and check the model cards. Also, we release the **[technical report](https://arxiv.org/abs/2309.16609)**. Please click the paper link and check it out!
In brief, we have strong base language models, which have been stably pretrained for up to 3 trillion tokens of multilingual data with a wide coverage of domains, languages (with a focus on Chinese and English), etc. They are able to achieve competitive performance on benchmark datasets. Additionally, we have chat models that are aligned with human preference based on SFT and RLHF (not released yet), which are able to chat, create content, extract information, summarize, translate, code, solve math problems, and so on, and are able to use tools, play as agents, or even play as code interpreters, etc.
* Information about Qwen for tool use, agent, and code interpreter
* Statistics of long-context understanding evaluation
* License agreement
* ...
Also, if you meet problems, turn to [FAQ](FAQ.md) for help first. Still feeling struggled? Feel free to shoot us issues (better in English so that more people can understand you)! If you would like to help us, send us pull requests with no hesitation! We are always excited about PR!
Would like to chat with us or date us coffee time? Welcome to our Discord or WeChat!
* 2023.11.30 🔥 We release **Qwen-72B** and **Qwen-72B-Chat**, which are trained on 3T tokens and support 32k context, along with **Qwen-1.8B**, and **Qwen-1.8B-Chat**, on ModelScope and Hugging Face. We have also strengthened the System Prompt capabilities of the Qwen-72B-Chat and Qwen-1.8B-Chat, see [example documentation](examples/system_prompt.md). Additionally, support the inference on **Ascend 910** and **Hygon DCU**. Check `ascend-support` and `dcu-support` for more details.
* 2023.9.25 🔥 We release **Qwen-14B** and **Qwen-14B-Chat** on ModelScope and Hugging Face, along with [qwen.cpp](https://github.com/QwenLM/qwen.cpp) and [Qwen-Agent](https://github.com/QwenLM/Qwen-Agent). Codes and checkpoints of **Qwen-7B** and **Qwen-7B-Chat** are also updated. **PLEASE PULL THE LATEST VERSION!**
- Compared to **Qwen-7B** (original), **Qwen-7B** uses more training tokens, increasing from 2.2T tokens to 2.4T tokens, while the context length extends from 2048 to 8192. The Chinese knowledge and coding ability of **Qwen-7B** have been further improved.
* 2023.8.21 We release the Int4 quantized model for Qwen-7B-Chat, **Qwen-7B-Chat-Int4**, which requires low memory costs but achieves improved inference speed. Besides, there is no significant performance degradation on the benchmark evaluation.
* 2023.8.3 We release both **Qwen-7B** and **Qwen-7B-Chat** on ModelScope and Hugging Face. We also provide a technical memo for more details about the model, including training details and model performance.
Qwen models outperform the baseline models of similar model sizes on a series of benchmark datasets, e.g., MMLU, C-Eval, GSM8K, MATH, HumanEval, MBPP, BBH, etc., which evaluate the models’ capabilities on natural language understanding, mathematic problem solving, coding, etc. Qwen-72B achieves better performance than LLaMA2-70B on all tasks and outperforms GPT-3.5 on 7 out of 10 tasks.
For all compared models, we report the best scores between their official reported results and [OpenCompass](https://opencompass.org.cn/leaderboard-llm).
For more experimental results (detailed model performance on more benchmark datasets) and details, please refer to our technical report by clicking [here](https://qianwen-res.oss-cn-beijing.aliyuncs.com/QWEN_TECHNICAL_REPORT.pdf).
You can use our pre-built docker images to skip most of the environment setup steps, see Section ["Using Pre-built Docker Images"](#-docker) for more details.
If not using docker, please make sure you have setup the environment and installed the required packages. Make sure you meet the above requirements, and then install the dependent libraries.
If your device supports fp16 or bf16, we recommend installing [flash-attention](https://github.com/Dao-AILab/flash-attention) (**we support flash attention 2 now.**) for higher efficiency and lower memory usage. (**flash-attention is optional and the project can run normally without installing it**)
To use Qwen-Chat for the inference, all you need to do is to input a few lines of codes as demonstrated below. Remember to pass in the correct model names or paths, such as "Qwen/Qwen-7B-Chat" and "Qwen/Qwen-14B-Chat". However, **please make sure that you are using the latest code.**
In the event of a network issue while attempting to download model checkpoints and codes from HuggingFace, an alternative approach is to initially fetch the checkpoint from ModelScope and then load it from the local directory as outlined below:
ModelScope is an open-source platform for Model-as-a-Service (MaaS), which provides flexible and cost-effective model service to AI developers. Similarly, you can run the models with ModelScope as shown below:
To deploy our models on CPU, we strongly advise you to use [qwen.cpp](https://github.com/QwenLM/qwen.cpp), which is a pure C++ implementation of Qwen and tiktoken. Check the repo for more details!
Also, it is also simple to directly run the model on CPU, which requires your specification of device:
```python
model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen-7B-Chat", device_map="cpu", trust_remote_code=True).eval()
```
However, it is likely that you suffer from extremely low inference efficiency.
### Multiple GPUs
If you suffer from lack of GPU memory and you would like to run the model on more than 1 GPU, you can directly use the default loading method, which is now supported by Transformers. The previous method based on `utils.py` is deprecated.
However, though this method is simple, the efficiency of the native pipeline parallelism is low. We advise you to use vLLM with FastChat and please read the section for deployment.
The most simple way to use Qwen through APIs is DashScope API service through Alibaba Cloud. We give an introduction to the usage. Additionally, we provide a script for you to deploy an OpenAI-style API on your own servers.
DashScope is the large language model API service provided by Alibaba Cloud, which now supports Qwen. Note that the models behind DashScope are in-house versions temporarily without details provided. The services include `qwen-turbo` and `qwen-plus`, where the former one runs faster and the latter achieves better performance. For more information, visit the documentation [here](https://dashscope.aliyun.com).
Please head to the official website [link](https://help.aliyun.com/zh/dashscope/developer-reference/activate-dashscope-and-create-an-api-key?spm=a2c4g.11186623.0.0.6c2774fahtfXdn) to create a DashScope account and obtain the API key (AK). We recommend setting the AK with an environment variable:
```bash
export DASHSCOPE_API_KEY="YOUR_DASHSCOPE_API_KEY"
```
Then please install the packages and click [here](https://help.aliyun.com/zh/dashscope/developer-reference/install-dashscope-sdk) for the documentation. If you use Python, you can install DashScope with pip:
```bash
pip install dashscope
```
If you use JAVA SDK, you can install it in this way:
We provide a solution based on [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), and release the Int4 and Int8 quantized models, which achieve nearly lossless model effects but improved performance on both memory costs and inference speed.
Here we demonstrate how to use our provided quantized models for inference. Before you start, make sure you meet the requirements of auto-gptq (e.g., torch 2.0 and above, transformers 4.32.0 and above, etc.) and install the required packages:
We illustrate the model performance of both BF16, Int8 and Int4 models on the benchmark, and we find that the quantized model does not suffer from significant performance degradation. Results are shown below:
The attention KV cache can be quantized and compressed for storage, to get a higher sample throughput. The arguments `use_cache_quantization` and `use_cache_kernel` in `config.json` are provided to enable KV cache quantization. The specific use method is as follows:
Attention: Currently, KV cache quantization and flash attention cannot be used at the same time.
If you enable KV cache quantization and flash attention at the same time (`use_flash_attn=True`, `use_cache_quantization=True`, `use_cache_kernel=True`), `use_flash_attn` is disabled by default (`use_flash_attn=false`).
We have verified that the use of the quantized Int8-KV-Cache model does not suffer from significant performance degradation in downstream evaluation. In the following, we focus on profiling its memory footprint in different conditions.
With KV cache quantization the model can save more memory when generating longer sequence (`sl`, sequence length, referring to the number of tokens generated) at the stage of inference.
The model with KV cache quantization will convert the format of `layer_past` from float to int8, and meanwhile the quantized `layer-past` will also store the quantization parameters.
If you want to use the attention KV which is quantized, you can use the dequantization operation to convert the Int8 key/value back to the float format as follows:
This section provides the statistics of speed and memory of models in different precisions. The speed and memory profiling are conducted using [this script](https://qianwen-res.oss-cn-beijing.aliyuncs.com/profile.py).
The profiling runs on a single A100-SXM4-80G GPU (except 2xA100 is mentioned) with PyTorch 2.0.1, CUDA 11.8, and Flash-Attention 2. (72B + vLLM uses PyTorch 2.1.0 and Cuda 11.8.) The inference speed is averaged over the encoded and generated tokens.
Note: The generation speed of the Int4/Int8 models mentioned above is provided by the autogptq library. The current speed of the model loaded using ``AutoModelForCausalLM.from_pretrained`` will be approximately 20% slower. We have reported this issue to the HuggingFace team and will update it promptly if a solution is available.
We also measure the inference speed and GPU memory usage with different settings of context and generation lengths, Flash-Attention version. You can find the results in the according modelcards on Hugging Face or ModelScope.
Now we provide the official training script, `finetune.py`, for users to finetune the pretrained model for downstream applications in a simple fashion. Additionally, we provide shell scripts to launch finetuning with no worries. This script supports the training with [DeepSpeed](https://github.com/microsoft/DeepSpeed) and [FSDP](https://engineering.fb.com/2021/07/15/open-source/fsdp/). The shell scripts that we provide use DeepSpeed (Note: this may have conflicts with the latest version of pydantic and you should use make sure `pydantic<2.0`) and Peft. You can install them by:
To prepare your training data, you need to put all the samples into a list and save it to a json file. Each sample is a dictionary consisting of an id and a list for conversation. Below is a simple example list with 1 sample:
Remember to specify the correct model name or path, the data path, as well as the output directory in the shell scripts. Another thing to notice is that we use DeepSpeed ZeRO 3 in this script. If you want to make changes, just remove the argument `--deepspeed` or make changes in the DeepSpeed configuration json file based on your requirements. Additionally, this script supports mixed-precision training, and thus you can use `--bf16 True` or `--fp16 True`. Remember to use DeepSpeed when you use fp16 due to mixed precision training. Empirically we advise you to use bf16 to make your training consistent with our pretraining and alignment if your machine supports bf16, and thus we use it by default.
Similarly, to run LoRA, use another script to run as shown below. Before you start, make sure that you have installed `peft`. Also, you need to specify your paths to your model, data, and output. We advise you to use absolute path for your pretrained model. This is because LoRA only saves the adapter and the absolute path in the adapter configuration json file is used for finding out the pretrained model to load. Also, this script support both bf16 and fp16.
In comparison with full-parameter finetuning, LoRA ([paper](https://arxiv.org/abs/2106.09685)) only updates the parameters of adapter layers but keeps the original large language model layers frozen. This allows much fewer memory costs and thus fewer computation costs.
Note that if you use LoRA to finetune the base language model, e.g., Qwen-7B, instead of chat models, e.g., Qwen-7B-Chat, the script automatically switches the embedding and output layer as trainable parameters. This is because the base language model has no knowledge of special tokens brought by ChatML format. Thus these layers should be updated for the model to understand and predict the tokens. Or in another word, if your training brings in special tokens in LoRA, you should set the layers to trainable parameters by setting `modules_to_save` inside the code. Also, if we have these parameters trainable, it is not available to use ZeRO 3, and this is why we use ZeRO 2 in the script by default. If you do not have new trainable parameters, you can switch to ZeRO 3 by changing the DeepSpeed configuration file. Additionally, we find that there is a significant gap between the memory footprint of LoRA with and without these trainable parameters. Therefore, if you have trouble with memory, we advise you to LoRA finetune the chat models. Check the profile below for more information.
If you still suffer from insufficient memory, you can consider Q-LoRA ([paper](https://arxiv.org/abs/2305.14314)), which uses the quantized large language model and other techniques such as paged attention to allow even fewer memory costs.
For Q-LoRA, we advise you to load our provided quantized model, e.g., Qwen-7B-Chat-Int4. You **SHOULD NOT** use the bf16 models. Different from full-parameter finetuning and LoRA, only fp16 is supported for Q-LoRA. For single-GPU training, we have to use DeepSpeed for mixed-precision training due to our observation of errors caused by torch amp. Besides, for Q-LoRA, the troubles with the special tokens in LoRA still exist. However, as we only provide the Int4 models for chat models, which means the language model has learned the special tokens of ChatML format, you have no worry about the layers. Note that the layers of the Int4 model should not be trainable, and thus if you introduce special tokens in your training, Q-LoRA might not work.
Different from full-parameter finetuning, the training of both LoRA and Q-LoRA only saves the adapter parameters. Suppose your training starts from Qwen-7B, you can load the finetuned model for inference as shown below:
If you want to merge the adapters and save the finetuned model as a standalone model (you can only do this with LoRA, and you CANNOT merge the parameters from Q-LoRA), you can run the following codes:
The `new_model_directory` directory will contain the merged model weights and module files. Please note that `*.cu` and `*.cpp` files may be missing in the saved files. If you wish to use the KV cache functionality, please manually copy them. Besides, the tokenizer files are not saved in the new directory in this step. You can copy the tokenizer files or use the following code
Note: For multi-GPU training, you need to specify the proper hyperparameters for distributed training based on your machine. Besides, we advise you to specify your maximum sequence length with the argument `--model_max_length`, based on your consideration of data, memory footprint, and training speed.
We profile the GPU memory and training speed of both LoRA (LoRA (emb) refers to training the embedding and output layer, while LoRA has no trainable embedding and output layer) and Q-LoRA in the setup of single-GPU training. In this test, we experiment on a single A100-SXM4-80G GPU, and we use CUDA 11.8 and Pytorch 2.0. Flash attention 2 is applied. We uniformly use a batch size of 1 and gradient accumulation of 8. We profile the memory (GB) and speed (s/iter) of inputs of different lengths, namely 256, 512, 1024, 2048, 4096, and 8192. We also report the statistics of full-parameter finetuning with Qwen-7B on 2 A100 GPUs. We only report the statistics of 256, 512, and 1024 tokens due to the limitation of GPU memory.
For Qwen-72B, we experiment in two ways: 1) Lora fintuning + DeepSpeed ZeRO 3 on 4 A100-SXM4-80G GPUs and 2) QLora (int4) fine-tuning on a single A100-SXM4-80G GPU. Note that OOM occurs on 4 A100-SXM4-80G GPUs both with LoRA (emb) fine-tuning and LoRA fine-tuning without Deepspeed ZeRO 3 (you can pass `--deepspeed finetune/ds_config_zero3.json` to [`finetune/finetune_lora_ds.sh`](finetune/finetune_lora_ds.sh) to enable DeepSpeed ZeRO 3).
Otherwise, please refer to the official vLLM [Installation Instructions](https://docs.vllm.ai/en/latest/getting_started/installation.html).
#### vLLM + Transformer-like Wrapper
You can download the [wrapper codes](examples/vllm_wrapper.py) and execute the following commands for multiple rounds of dialogue interaction. (Note: It currently only supports the ``model.chat()`` method.)
```python
from vllm_wrapper import vLLMWrapper
model = vLLMWrapper('Qwen/Qwen-7B-Chat', tensor_parallel_size=1)
response, history = model.chat(query="你好", history=None)
print(response)
response, history = model.chat(query="给我讲一个年轻人奋斗创业最终取得成功的故事。", history=history)
print(response)
response, history = model.chat(query="给这个故事起一个标题", history=history)
print(response)
```
#### vLLM + Web Demo / OpenAI-like API
You can use FastChat to lauch a web demo or an OpenAI API server. First, install FastChat:
However, if you hope to run the model on multiple GPUs for faster inference or larger memory, you can use tensor parallelism supported by vLLM. Suppose you run the model on 4 GPUs, the command is shown below:
We provide a CLI demo example in `cli_demo.py`, which supports streaming output for the generation. Users can interact with Qwen-7B-Chat by inputting prompts, and the model returns model outputs in the streaming mode. Run the command below:
You can change your arguments, e.g., `-c` for checkpoint name or path, `--cpu-only` for CPU deployment, etc. If you meet problems launching your API deployment, updating the packages to the latest version can probably solve them.
Using the API is also simple. See the example below:
**Function calling** is also supported (but only when `stream=False` for the moment). See the [example usage](examples/function_call_examples.py) here.
To simplify the deployment process, we provide docker images with pre-built environments: [qwenllm/qwen](https://hub.docker.com/r/qwenllm/qwen). You only need to install the driver and download model files to launch demos, deploy OpenAI API, and finetune the model.
### Preparation
1. Install the correct version of Nvidia driver depending on the image to use:
-`qwenllm/qwen:latest`: same as `qwenllm/qwen:cu117`
2. Install and configure [docker](https://docs.docker.com/engine/install/) and [nvidia-container-toolkit](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html):
The commands above will automatically download the required image and launch a Web UI demo in background (the service will auto-restart). You can open `http://localhost:${PORT}` on the host to use the demo.
The demo is successfully launched if you see the following output:
```text
Successfully started web demo. Open '...' to try!
Run `docker logs ...` to check demo status.
Run `docker rm -f ...` to stop and remove the demo.
```
If you want to check the status of the demo, you can use `docker logs qwen` to display outputs.
You can use `docker rm -f qwen` to stop the service and remove the container.
### Finetuning
The method of finetuning using the pre-built Docker image is basically the same as [the above chapter](#Finetuning) (we have already installed dependencies in the image):
The following is an example of single-GPU LoRA:
```bash
IMAGE_NAME=qwenllm/qwen:cu117
CHECKPOINT_PATH=/path/to/Qwen-7B # Path to downloaded model checkpoints and codes
#CHECKPOINT_PATH=/path/to/Qwen-7B-Chat-Int4 # Path to downloaded model checkpoints and codes (Q-LoRA)
DATA_PATH=/path/to/data/root # Prepare finetune data at ${DATA_PATH}/example.json
OUTPUT_PATH=/path/to/output/checkpoint # Path to finetune outputs
# Use all host devices by default
DEVICE=all
# If you need to specify GPUs for training, set device as follow (NOTE: internal quotation marks cannot be omitted)
Qwen-1.8-Chat and Qwen-72B-Chat have been fully trained on diverse system prompts with multiple rounds of complex interactions, so that they can follow a variety of system prompts and realize model customization in context, further improving the scalability of Qwen-chat.
With System Prompt, Qwen-Chat can realize **roly playing**, **language style transfer**, **task setting**, and **behavior setting**.
For more information, please refer to the [example documentation](examples/system_prompt.md).
Qwen-Chat has been optimized for tool usage and function calling capabilities. Users can develop agents, LangChain applications, and even augment Qwen with a Python Code Interpreter.
We provide documentation on how to implement tool calls based on the principle of ReAct Prompting, please refer to [the ReAct example](examples/react_prompt.md). Based on this principle, we provide support for function calling in [openai_api.py](openai_api.py).
We have tested the model's tool calling capabilities on our open-source Chinese evaluation benchmark and found that Qwen-Chat consistently performs well:
To assess Qwen's ability to use the Python Code Interpreter for tasks such as mathematical problem solving, data visualization, and other general-purpose tasks such as file handling and web scraping, we have created and open-sourced a benchmark specifically designed for evaluating these capabilities. You can find the benchmark at this [link](https://github.com/QwenLM/Qwen-Agent/tree/main/benchmark).
To extend the context length and break the bottleneck of training sequence length, we introduce several techniques, including NTK-aware interpolation, window attention, and LogN attention scaling, to extend the context length of Qwen-14B from 2K to over 8K tokens, and Qwen-1.8B/7B from 8K to 32K tokens.
For Qwen-72B, we adapt RoPE to longer contexts with a larger rotary base. Qwen-72B supports the max context length of 32K tokens.
We conduct language modeling experiments on the arXiv dataset with the PPL evaluation and find that Qwen can reach outstanding performance in the scenario of long context. Results are demonstrated below:
Furthermore, to verify the ability of Qwen-72B-Chat on long text understanding, we tested it on [L-Eval](https://arxiv.org/abs/2307.11088) (closed-ended tasks). The results are as follows:
| Model | Input Length | Average | Coursera | GSM | QuALITY | TOEFL | CodeU | SFcition |
We conducted the "needle in a haystack" experiment (the idea came from [@Greg Kamradt](https://twitter.com/GregKamradt/status/1727018183608193393)) to test whether the model can retrieve information at different positions in the inputs of different lengths, the result is as follows:
The above results show that Qwen-72B-Chat can accurately retrieve information placed in various positions within an input length of 32k, proving its excellent long text understanding capabilities.
Our tokenizer based on tiktoken is different from other tokenizers, e.g., sentencepiece tokenizer. You need to pay attention to special tokens, especially in finetuning. For more detailed information on the tokenizer and related use in fine-tuning, please refer to the [documentation](tokenization_note.md).
For your reproduction of the model performance on benchmark datasets, we provide scripts for you to reproduce the results. Check [eval/EVALUATION.md](eval/EVALUATION.md) for more information. Note that the reproduction may lead to slight differences from our reported results.
author={Jinze Bai and Shuai Bai and Yunfei Chu and Zeyu Cui and Kai Dang and Xiaodong Deng and Yang Fan and Wenbin Ge and Yu Han and Fei Huang and Binyuan Hui and Luo Ji and Mei Li and Junyang Lin and Runji Lin and Dayiheng Liu and Gao Liu and Chengqiang Lu and Keming Lu and Jianxin Ma and Rui Men and Xingzhang Ren and Xuancheng Ren and Chuanqi Tan and Sinan Tan and Jianhong Tu and Peng Wang and Shijie Wang and Wei Wang and Shengguang Wu and Benfeng Xu and Jin Xu and An Yang and Hao Yang and Jian Yang and Shusheng Yang and Yang Yao and Bowen Yu and Hongyi Yuan and Zheng Yuan and Jianwei Zhang and Xingxuan Zhang and Yichang Zhang and Zhenru Zhang and Chang Zhou and Jingren Zhou and Xiaohuan Zhou and Tianhang Zhu},
The source code provided at <https://github.com/QwenLM/Qwen> is licensed under the [Apache 2.0 License](./LICENSE) that can be found at the root directory.
Researchers and developers are free to use the codes and model weights of both Qwen and Qwen-Chat. For their commercial use, please check the License Agreement accompanying each model.
- Qwen-72B, Qwen-14B, and Qwen-7B are licensed under the [Tongyi Qianwen LICENSE AGREEMENT](./Tongyi%20Qianwen%20LICENSE%20AGREEMENT) that can be found at the corresponding HuggingFace and ModelScope repository. For commercial use, please fill out the form ([72B](https://dashscope.console.aliyun.com/openModelApply/Qwen-72B-Chat), [14B](https://dashscope.console.aliyun.com/openModelApply/Qwen-14B-Chat), and [7B](https://dashscope.console.aliyun.com/openModelApply/qianwen)) to apply.
- Qwen-1.8B is licensed under the [Tongyi Qianwen RESEARCH LICENSE AGREEMENT](./Tongyi%20Qianwen%20RESEARCH%20LICENSE%20AGREEMENT) that can be found at the corresponding HuggingFace and ModelScope repository. For commercial use, please contact us.
If you are interested to leave a message to either our research team or product team, join our Discord or WeChat groups! Also, feel free to send an email to qianwen_opensource@alibabacloud.com.
